• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion

Incremental Backup size increases

Schlaubi

New Pleskian
Username:

TITLE

Incremental Backup size increases

PRODUCT, VERSION, OPERATING SYSTEM, ARCHITECTURE

OS: Ubuntu 18.04.5 LTS, Plesk Obsidian 18.0.39
OS: Ubuntu 20 LTS, Plesk Obsidian 18.0.39 (fallback server, with no user traffic or content change at all.)

PROBLEM DESCRIPTION

Screenshots are shown at Issue - Incremental Backup size increases

We discovered that plesk backups are continuously increasing and consuming more and more diskspace despite the domain content rarely changes (except expected ~10mb MariaDB). Plesk's "Backup Manager" shows an increase of ~330mb of two incremental backups within 2 days. These diffs are shown in comp_Inc1_Inc2.jpg (left Day1, right Day2). It seems like non domain content is getting included in every incremental backup again. Purple entries are new entries, black are equal ones.
We looked a bit deeper into backup_ext_nodejs_*.tgz. Incremental backup from Day2 (right) includes the almost same backup_ext_nodejs_*.tgz from Day1 again. backup_ext_nodejs_Day2.jpg shows a hex compare with almost no diff of the Day1 tgz compared to Day2 tgz.

It looks like a bug, where the same content is included repeatedly leading to increased backup sizes. We would expect only small incremental backups.

In the provided screenshot comp_Inc1_Inc2.jpg you can see, that incremental backup from Day2 (right) includes the same file backup_ext_nodejs_*.tgz 3 times with 171mb each. All these files (purple ones) are included in every incremental backup despite they don't change.
Screenshot backup_ext_nodejs_day2_7zip.jpg shows both backup_ext_nodejs_*.tgz from Day1 and Day2. CRC of both files is the same (green). 3 different bytes as shown in previous backup_ext_nodejs_Day2 seems to be the different changed date (red). Content of these 171mb backups is binary equal.

That situation leads to increasing backup sizes (as seen in comp_Inc1_Inc2.jpg).
Full backup day0
backup_ext_nodejs_2111280005.tgz (170mb)

Incremental day1
backup_ext_nodejs_2111280005.tgz (170mb)
backup_ext_nodejs_2111280005_2111290005.tgz (+170mb)

Incremental day2
backup_ext_nodejs_2111280005.tgz (170mb)
backup_ext_nodejs_2111280005_2111290005.tgz (170mb)
backup_ext_nodejs_2111280005_2111300005.tgz (+170mb)
etc.

As an immediate workaround to protect our servers from running out of free disk space we had to drastically reduce backup creations. We're not happy with that situation in a productional environment.

STEPS TO REPRODUCE

- Default Plesk installation on Linux
- Wait until several scheduled backups

ACTUAL RESULT

- daily incremental backup size ~330mb

EXPECTED RESULT

- daily incremental backup size ~20mb

ANY ADDITIONAL INFORMATION

(DID NOT ANSWER QUESTION)

YOUR EXPECTATIONS FROM PLESK SERVICE TEAM

Confirm bug
 
Those suffering from PLESK's incremental backups should see about writing a script using TAR's incremental backups.
Mine has saved my fanny many times!
 
Back
Top